MadSystems Seminar -- Jing Liu (Microsoft)
Event Details
Date
Tuesday, February 3, 2026
Time
4-5 p.m.
Location
6618 Morgridge Hall
Description
- Talk Title: Towards Instance-Customizable Operating Systems
- Abstract: My research explores instance-customizable operating systems—systems that can be safely customized at runtime to hardware, workloads, and failure contexts without breaking compatibility. This talk covers my work on customizing OS I/O subsystem structure (uFS), Ananke, which enables OS-level customization of failure semantics, and KernelX, which supports safe, in-situ tuning of performance-critical constants in deployed kernels, achieving millisecond-scale updates and significant performance gains through a programmable tuning policy plane in user space. Together, these systems illustrate how principled OS abstractions can expose rich optimization opportunities that AI techniques can meaningfully manage.
- Speaker Bio: Jing Liu is a Senior Researcher at Microsoft Research Asia. She received her Ph.D. from the University of Wisconsin–Madison in 2024. Her research includes datacenter-scale operating systems, file and storage systems, and distributed systems, with a focus on redesigning system mechanisms and interfaces so that machine learning–based techniques can meaningfully participate in system problems such as adaptation, control, and optimization. Jing is a recipient of the Erik Riedel Best Paper Award at FAST 2025 and a Meta Ph.D. Research Fellowship in Systems. She has served on the program committees of SOSP, OSDI, NSDI, and FAST, and is a Shadow PC Co-Chair for EuroSys 2026. More details can be found at https://jingliu.xyz.
Cost
Free
Contact
Accessibility
We value inclusion and access for all participants and are pleased to provide reasonable accommodations for this event. Please call 608-867-6867 or email tomy.1516@gmail.com to make a disability-related accommodation request. Reasonable effort will be made to support your request.